/// <summary>
/// Read macros in the MMF file.
/// </summary>
/// <param name="filePath">MMF file.</param>
/// <returns>Macros.</returns>
private static IEnumerable<Macro> Macros(string filePath)
{
Helper.ThrowIfNull(filePath);
using (HmmReader reader = new HmmReader(filePath))
{
for (char macroType = ' '; reader.ReadNextMacro(out macroType);)
{
Macro macro = new Macro() { Type = macroType, Reader = reader };
yield return macro;
if (macroType == 'h')
{
break;
}
}
}
}
/// <summary>
/// Get state distribution and physical HMM model mapping [Key = "state name", Value = "HMM names"].
/// </summary>
/// <param name="fileName">Binary HMM model file.</param>
/// <param name="stateTag">Tag indicating which model mapping to get.</param>
/// <returns>The state distribution->model mapping.</returns>
public static Dictionary<string, Collection<string>> GetStateAndHmmMapping(string fileName, string stateTag)
{
Helper.ThrowIfNull(fileName);
Helper.ThrowIfNull(stateTag);
Dictionary<string, Collection<string>> mapping = new Dictionary<string, Collection<string>>();
using (HmmReader hr = new HmmReader(fileName))
{
bool isHmmPart = false;
string hmmName = string.Empty;
char macroType = ' ';
while (hr.ReadNextMacro(out macroType))
{
if (macroType.Equals('p') || macroType.Equals('s'))
{
string state = hr.ReadString();
if (isHmmPart)
{
if (mapping.ContainsKey(state))
{
mapping[state].Add(hmmName);
}
}
else
{
// check whether it's the required type
if (state.Contains(stateTag))
{
mapping.Add(state, new Collection<string>());
}
// Don't skip reading the Gaussian distribution!!
hr.ReadGaussian();
}
}
else if (macroType.Equals('t'))
{
if (!isHmmPart)
{
hr.ReadTransMatrix();
}
}
else if (macroType.Equals('h'))
{
hmmName = hr.ReadString();
if (!isHmmPart)
{
isHmmPart = true;
}
}
else if (macroType == 'v')
{
hr.ReadVariance();
}
}
}
return mapping;
}
/// <summary>
/// Get physical HMM and state distribution model mapping [Key = "HMM name",
/// Value = "model names". First dimension is state, second dimension is stream].
/// </summary>
/// <param name="mmfFileName">Binary HMM model file.</param>
/// <returns>The model->state distribution mapping.</returns>
public static Dictionary<string, string[,]> GetHmmAndStateMapping(string mmfFileName)
{
Helper.ThrowIfNull(mmfFileName);
Dictionary<string, Collection<string>> hmmModels = new Dictionary<string, Collection<string>>();
int streamNumber = -1;
using (HmmReader hr = new HmmReader(mmfFileName))
{
bool isHmmPart = false;
string hmmName = string.Empty;
char macroType = ' ';
while (hr.ReadNextMacro(out macroType))
{
if (macroType.Equals('p') || macroType.Equals('s'))
{
string state = hr.ReadString();
if (isHmmPart)
{
Debug.Assert(hmmModels.ContainsKey(hmmName));
hmmModels[hmmName].Add(state);
}
else
{
// Don't skip reading the Gaussian distribution!!
hr.ReadGaussian();
}
}
else if (macroType.Equals('t'))
{
if (!isHmmPart)
{
hr.ReadTransMatrix();
}
}
else if (macroType.Equals('h'))
{
hmmName = hr.ReadString();
if (!hmmModels.ContainsKey(hmmName))
{
hmmModels.Add(hmmName, new Collection<string>());
}
if (!isHmmPart)
{
isHmmPart = true;
}
}
else if (macroType.Equals('o'))
{
HmmSymbol symbol = hr.ReadNextSymbol();
Debug.Assert(symbol == HmmSymbol.StreamInfo);
short[] mmfStreamWidths = hr.ReadInt16Array();
streamNumber = mmfStreamWidths.Length;
}
else if (macroType == 'v')
{
hr.ReadVariance();
}
}
}
Dictionary<string, string[,]> model2Streams = new Dictionary<string, string[,]>();
foreach (KeyValuePair<string, Collection<string>> pair in hmmModels)
{
Debug.Assert(pair.Value.Count % streamNumber == 0);
int stateNumber = pair.Value.Count / streamNumber;
string[,] streams = new string[stateNumber, streamNumber];
for (int i = 0; i < stateNumber; ++i)
{
for (int j = 0; j < streamNumber; ++j)
{
streams[i, j] = pair.Value[(i * streamNumber) + j];
}
}
model2Streams.Add(pair.Key, streams);
}
return model2Streams;
}
/// <summary>
/// Get Gaussians width information.
/// </summary>
/// <param name="hmr">Model reader.</param>
/// <param name="streamIndexes">Stream indexes.</param>
/// <returns>Stream Widths.</returns>
public static int[] ReadStreamWidths(HmmReader hmr, int[] streamIndexes)
{
Helper.ThrowIfNull(hmr);
Helper.ThrowIfNull(streamIndexes);
HmmSymbol symbol = hmr.ReadNextSymbol();
Debug.Assert(symbol == HmmSymbol.StreamInfo);
short[] mmfStreamWidths = hmr.ReadInt16Array();
Debug.Assert(streamIndexes.Length <= mmfStreamWidths.Length);
int[] streamWidths = new int[streamIndexes.Length];
for (int i = 0; i < streamIndexes.Length; i++)
{
streamWidths[i] = mmfStreamWidths[streamIndexes[i] - 1]; // To 0 based index
}
if (hmr.ReadNextSymbol() == HmmSymbol.MSDINFO)
{
hmr.ReadInt16Array();
}
return streamWidths;
}